Cogestion des ressources naturelles : une approche structurale pour quantifier la contribution des réseaux d'acteurs à la résilience des systèmes socio-écologiques

Alors que les activités anthropiques font basculer de nombreux écosystèmes vers des régimes fonctionnels différents, la résilience des systèmes socio-écologiques devient un problème pressant. Des acteurs locaux, impliqués dans une grande diversité de groupes — allant d’initiatives locales et indépendantes à de grandes institutions formelles — peuvent agir sur ces questions en collaborant au développement, à la promotion ou à l’implantation de pratiques plus en accord avec ce que l’environnement peut fournir. De ces collaborations répétées émergent des réseaux complexes, et il a été montré que la topologie de ces réseaux peut améliorer la résilience des systèmes socio-écologiques (SSÉ) auxquels ils participent. La topologie des réseaux d’acteurs favorisant la résilience de leur SSÉ est caractérisée par une combinaison de plusieurs facteurs : la structure doit être modulaire afin d’aider les différents groupes à développer et proposer des solutions à la fois plus innovantes (en réduisant l’homogénéisation du réseau), et plus proches de leurs intérêts propres ; elle doit être bien connectée et facilement synchronisable afin de faciliter les consensus, d’augmenter le capital social, ainsi que la capacité d’apprentissage ; enfin, elle doit être robuste, afin d’éviter que les deux premières caractéristiques ne souffrent du retrait volontaire ou de la mise à l’écart de certains acteurs. Ces caractéristiques, qui sont relativement intuitives à la fois conceptuellement et dans leur application mathématique, sont souvent employées séparément pour analyser les qualités structurales de réseaux d’acteurs empiriques. Cependant, certaines sont, par nature, incompatibles entre elles. Par exemple, le degré de modularité d’un réseau ne peut pas augmenter au même rythme que sa connectivité, et cette dernière ne peut pas être améliorée tout en améliorant sa robustesse. Cet obstacle rend difficile la création d’une mesure globale, car le niveau auquel le réseau des acteurs contribue à améliorer la résilience du SSÉ ne peut pas être la simple addition des caractéristiques citées, mais plutôt le résultat d’un compromis subtil entre celles-ci. Le travail présenté ici a pour objectifs (1), d’explorer les compromis entre ces caractéristiques ; (2) de proposer une mesure du degré auquel un réseau empirique d’acteurs contribue à la résilience de son SSÉ ; et (3) d’analyser un réseau empirique à la lumière, entre autres, de ces qualités structurales. Cette thèse s’articule autour d’une introduction et de quatre chapitres numérotés de 2 à 5. Le chapitre 2 est une revue de la littérature sur la résilience des SSÉ. Il identifie une série de caractéristiques structurales (ainsi que les mesures de réseaux qui leur correspondent) liées à l’amélioration de la résilience dans les SSÉ. Le chapitre 3 est une étude de cas sur la péninsule d’Eyre, une région rurale d’Australie-Méridionale où l’occupation du sol, ainsi que les changements climatiques, contribuent à l’érosion de la biodiversité. Pour cette étude de cas, des travaux de terrain ont été effectués en 2010 et 2011 durant lesquels une série d’entrevues a permis de créer une liste des acteurs de la cogestion de la biodiversité sur la péninsule. Les données collectées ont été utilisées pour le développement d’un questionnaire en ligne permettant de documenter les interactions entre ces acteurs. Ces deux étapes ont permis la reconstitution d’un réseau pondéré et dirigé de 129 acteurs individuels et 1180 relations. Le chapitre 4 décrit une méthodologie pour mesurer le degré auquel un réseau d’acteurs participe à la résilience du SSÉ dans lequel il est inclus. La méthode s’articule en deux étapes : premièrement, un algorithme d’optimisation (recuit simulé) est utilisé pour fabriquer un archétype semi-aléatoire correspondant à un compromis entre des niveaux élevés de modularité, de connectivité et de robustesse. Deuxièmement, un réseau empirique (comme celui de la péninsule d’Eyre) est comparé au réseau archétypique par le biais d’une mesure de distance structurelle. Plus la distance est courte, et plus le réseau empirique est proche de sa configuration optimale. La cinquième et dernier chapitre est une amélioration de l’algorithme de recuit simulé utilisé dans le chapitre 4. Comme il est d’usage pour ce genre d’algorithmes, le recuit simulé utilisé projetait les dimensions du problème multiobjectif dans une seule dimension (sous la forme d’une moyenne pondérée). Si cette technique donne de très bons résultats ponctuellement, elle n’autorise la production que d’une seule solution parmi la multitude de compromis possibles entre les différents objectifs. Afin de mieux explorer ces compromis, nous proposons un algorithme de recuit simulé multiobjectifs qui, plutôt que d’optimiser une seule solution, optimise une surface multidimensionnelle de solutions. Cette étude, qui se concentre sur la partie sociale des systèmes socio-écologiques, améliore notre compréhension des structures actorielles qui contribuent à la résilience des SSÉ. Elle montre que si certaines caractéristiques profitables à la résilience sont incompatibles (modularité et connectivité, ou — dans une moindre mesure — connectivité et robustesse), d’autres sont plus facilement conciliables (connectivité et synchronisabilité, ou — dans une moindre mesure — modularité et robustesse). Elle fournit également une méthode intuitive pour mesurer quantitativement des réseaux d’acteurs empiriques, et ouvre ainsi la voie vers, par exemple, des comparaisons d’études de cas, ou des suivis — dans le temps — de réseaux d’acteurs. De plus, cette thèse inclut une étude de cas qui fait la lumière sur l’importance de certains groupes institutionnels pour la coordination des collaborations et des échanges de connaissances entre des acteurs aux intérêts potentiellement divergents.

Optimization of Cooling Protocols for Hearts Destined for Transplantation

Design and analysis of conceptually different cooling systems for the human heart preservation are numerically investigated. A heart cooling container with required connections was designed for a normal size human heart. A three-dimensional, high resolution human heart geometric model obtained from CT-angio data was used for simulations. Nine different cooling designs are introduced in this research. The first cooling design (Case 1) used a cooling gelatin only outside of the heart. In the second cooling design (Case 2), the internal parts of the heart were cooled via pumping a cooling liquid inside both the heart’s pulmonary and systemic circulation systems. An unsteady conjugate heat transfer analysis is performed to simulate the temperature field variations within the heart during the cooling process. Case 3 simulated the currently used cooling method in which the coolant is stagnant. Case 4 was a combination of Case 1 and Case 2. A linear thermoelasticity analysis was performed to assess the stresses applied on the heart during the cooling process. In Cases 5 through 9, the coolant solution was used for both internal and external cooling. For external circulation in Case 5 and Case 6, two inlets and two outlets were designed on the walls of the cooling container. Case 5 used laminar flows for coolant circulations inside and outside of the heart. Effects of turbulent flow on cooling of the heart were studied in Case 6. In Case 7, an additional inlet was designed on the cooling container wall to create a jet impinging the hot region of the heart’s wall. Unsteady periodic inlet velocities were applied in Case 8 and Case 9. The average temperature of the heart in Case 5 was +5.0oC after 1500 s of cooling. Multi-objective constrained optimization was performed for Case 5. Inlet velocities for two internal and one external coolant circulations were the three design variables for optimization. Minimizing the average temperature of the heart, wall shear stress and total volumetric flow rates were the three objectives. The only constraint was to keep von Mises stress below the ultimate tensile stress of the heart’s tissue.

Effective end wall profiling rules for a highly loaded compressor cascade

This paper presents a numerical study of a linear compressor cascade to investigate the effective end wall profiling rules for highly-loaded axial compressors. The first step in the research applies a correlation analysis for the different flow field parameters by a data mining over 600 profiling samples to quantify how variations of loss, secondary flow and passage vortex interact with each other under the influence of a profiled end wall. The result identifies the dominant role of corner separation for control of total pressure loss, providing a principle that only in the flow field with serious corner separation does the does the profiled end wall change total pressure loss, secondary flow and passage vortex in the same direction. Then in the second step, a multi-objective optimization of a profiled end wall is performed to reduce loss at design point and near stall point. The development of effective end wall profiling rules is based on the manner of secondary flow control rather than the geometry features of the end wall. Using the optimum end wall cases from the Pareto front, a quantitative tool for analyzing secondary flow control is employed. The driving force induced by a profiled end wall on different regions of end wall flow are subjected to a detailed analysis and identified for their positive/negative influences in relieving corner separation, from which the effective profiling rules are further confirmed. It is found that the profiling rules on a cascade show distinct differences at design point and near stall point, thus loss control of different operating points is generally independent.

HNF1B variants associate with promoter methylation and regulate gene networks activated in prostate and ovarian cancer

Two independent regions within HNF1B are consistently identified in prostate and ovarian cancer genome-wide association studies (GWAS); their functional roles are unclear. We link prostate cancer (PC) risk SNPs rs11649743 and rs3760511 with elevated HNF1B gene expression and allele-specific epigenetic silencing, and outline a mechanism by which common risk variants could effect functional changes that increase disease risk: functional assays suggest that HNF1B is a pro-differentiation factor that suppresses epithelial-to-mesenchymal transition (EMT) in unmethylated, healthy tissues. This tumor-suppressor activity is lost when HNF1B is silenced by promoter methylation in the progression to PC. Epigenetic inactivation of HNF1B in ovarian cancer also associates with known risk SNPs, with a similar impact on EMT. This represents one of the first comprehensive studies into the pleiotropic role of a GWAS-associated transcription factor across distinct cancer types, and is the first to describe a conserved role for a multi-cancer genetic risk factor.

Management of public water supply to reduce energy cost and improve wind power uptake

This paper presents a study on the implementation of Real-Time Pricing (RTP) based Demand Side Management (DSM) of water pumping at a clean water pumping station in Northern Ireland, with the intention of minimising electricity costs and maximising the usage of electricity from wind generation. A Genetic Algorithm (GA) was used to create pumping schedules based on system constraints and electricity tariff scenarios. Implementation of this method would allow the water network operator to make significant savings on electricity costs while also helping to mitigate the variability of wind generation.

Management of public water supply to reduce energy cost and improve wind power uptake

This paper presents a study on the implementation of Real-Time Pricing (RTP) based Demand Side Management (DSM) of water pumping at a clean water pumping station in Northern Ireland, with the intention of minimising electricity costs and maximising the usage of electricity from wind generation. A Genetic Algorithm (GA) was used to create pumping schedules based on system constraints and electricity tariff scenarios. Implementation of this method would allow the water network operator to make significant savings on electricity costs while also helping to mitigate the variability of wind generation.

Contrôle adaptatif d'un bioréacteur cardiaque par algorithme génétique

Un bon fonctionnement du coeur humain est primordial pour maintenir une bonne qualité de vie. Cependant, lorsque le coeur est défaillant, certaines interventions chirurgicales s’avèrent nécessaires pour prolonger l’espérance de vie. Dans le cadre d’un projet multidisciplinaire reliant le génie mécanique avec le domaine biomédical, notre équipe travaille sur la fabrication de valves cardiaques conçues entièrement par génie tissulaire. Pour y parvenir, il est important d’obtenir des propriétés mécaniques optimales pour les tissus biologiques. Afin d’obtenir ces propriétés mécaniques, un outil important a été fabriqué lors d’une étude antérieure : le bioréacteur cardiaque. Le bioréacteur cardiaque permet de reproduire l’environnement physiologique du coeur, notamment les conditions de débit et de pression. Il est crucial de bien contrôler ces conditions, car celles-ci jouent un rôle important lors du conditionnement des substituts valvulaires. Toutefois, il est complexe de contrôler simultanément ces deux conditions de manière efficace. C’est pourquoi notre équipe s’est concentrée sur le développement d’une nouvelle stratégie de contrôle afin que le bioréacteur puisse reproduire le plus fidèlement possible l’environnement physiologique. Plusieurs techniques de contrôle ont été essayés jusqu’à maintenant. Par contre, leur précision était généralement limitée. Une nouvelle approche a donc été envisagée et est présentée dans ce mémoire. Cette nouvelle approche pour le contrôle du bioréacteur est basée sur un type d’algorithme bien connu mais encore très peu utilisé en contrôle : les algorithmes génétiques. Cette approche prometteuse nous a permis de produire des résultats dépassant tous ceux obtenus jusqu’à maintenant pour l’une des deux conditions, soit le débit physiologique.

Predictive Analytics and Decision Support for Heart Failure patients

Thesis (Master's)--University of Washington, 2016-08

'Bayesian Optimisation for Nurse Scheduling'

A Bayesian optimisation algorithm for a nurse scheduling problem is presented, which involves choosing a suitable scheduling rule from a set for each nurse's assignment. When a human scheduler works, he normally builds a schedule systematically following a set of rules. After much practice, the scheduler gradually masters the knowledge of which solution parts go well with others. He can identify good parts and is aware of the solution quality even if the scheduling process is not yet completed, thus having the ability to finish a schedule by using flexible, rather than fixed, rules. In this paper, we design a more human-like scheduling algorithm, by using a Bayesian optimisation algorithm to implement explicit learning from past solutions. A nurse scheduling problem from a UK hospital is used for testing. Unlike our previous work that used Genetic Algorithms to implement implicit learning [1], the learning in the proposed algorithm is explicit, i.e. we identify and mix building blocks directly. The Bayesian optimisation algorithm is applied to implement such explicit learning by building a Bayesian network of the joint distribution of solutions. The conditional probability of each variable in the network is computed according to an initial set of promising solutions. Subsequently, each new instance for each variable is generated by using the corresponding conditional probabilities, until all variables have been generated, i.e. in our case, new rule strings have been obtained. Sets of rule strings are generated in this way, some of which will replace previous strings based on fitness. If stopping conditions are not met, the conditional probabilities for all nodes in the Bayesian network are updated again using the current set of promising rule strings. For clarity, consider the following toy example of scheduling five nurses with two rules (1: random allocation, 2: allocate nurse to low-cost shifts). In the beginning of the search, the probabilities of choosing rule 1 or 2 for each nurse is equal, i.e. 50%. After a few iterations, due to the selection pressure and reinforcement learning, we experience two solution pathways: Because pure low-cost or random allocation produces low quality solutions, either rule 1 is used for the first 2-3 nurses and rule 2 on remainder or vice versa. In essence, Bayesian network learns 'use rule 2 after 2-3x using rule 1' or vice versa. It should be noted that for our and most other scheduling problems, the structure of the network model is known and all variables are fully observed. In this case, the goal of learning is to find the rule values that maximize the likelihood of the training data. Thus, learning can amount to 'counting' in the case of multinomial distributions. For our problem, we use our rules: Random, Cheapest Cost, Best Cover and Balance of Cost and Cover. In more detail, the steps of our Bayesian optimisation algorithm for nurse scheduling are: 1. Set t = 0, and generate an initial population P(0) at random; 2. Use roulette-wheel selection to choose a set of promising rule strings S(t) from P(t); 3. Compute conditional probabilities of each node according to this set of promising solutions; 4. Assign each nurse using roulette-wheel selection based on the rules' conditional probabilities. A set of new rule strings O(t) will be generated in this way; 5. Create a new population P(t+1) by replacing some rule strings from P(t) with O(t), and set t = t+1; 6. If the termination conditions are not met (we use 2000 generations), go to step 2. Computational results from 52 real data instances demonstrate the success of this approach. They also suggest that the learning mechanism in the proposed approach might be suitable for other scheduling problems. Another direction for further research is to see if there is a good constructing sequence for individual data instances, given a fixed nurse scheduling order. If so, the good patterns could be recognized and then extracted as new domain knowledge. Thus, by using this extracted knowledge, we can assign specific rules to the corresponding nurses beforehand, and only schedule the remaining nurses with all available rules, making it possible to reduce the solution space. Acknowledgements The work was funded by the UK Government's major funding agency, Engineering and Physical Sciences Research Council (EPSRC), under grand GR/R92899/01. References [1] Aickelin U, "An Indirect Genetic Algorithm for Set Covering Problems", Journal of the Operational Research Society, 53(10): 1118-1126,

'The application of Bayesian Optimization and Classifier Systems in Nurse Scheduling'

Abstract. Two ideas taken from Bayesian optimization and classifier systems are presented for personnel scheduling based on choosing a suitable scheduling rule from a set for each person's assignment. Unlike our previous work of using genetic algorithms whose learning is implicit, the learning in both approaches is explicit, i.e. we are able to identify building blocks directly. To achieve this target, the Bayesian optimization algorithm builds a Bayesian network of the joint probability distribution of the rules used to construct solutions, while the adapted classifier system assigns each rule a strength value that is constantly updated according to its usefulness in the current situation. Computational results from 52 real data instances of nurse scheduling demonstrate the success of both approaches. It is also suggested that the learning mechanism in the proposed approaches might be suitable for other scheduling problems.

An Estimation of Distribution Algorithm for Nurse Scheduling

Schedules can be built in a similar way to a human scheduler by using a set of rules that involve domain knowledge. This paper presents an Estimation of Distribution Algorithm (EDA) for the nurse scheduling problem, which involves choosing a suitable scheduling rule from a set for the assignment of each nurse. Unlike previous work that used Genetic Algorithms (GAs) to implement implicit learning, the learning in the proposed algorithm is explicit, i.e. we identify and mix building blocks directly. The EDA is applied to implement such explicit learning by building a Bayesian network of the joint distribution of solutions. The conditional probability of each variable in the network is computed according to an initial set of promising solutions. Subsequently, each new instance for each variable is generated by using the corresponding conditional probabilities, until all variables have been generated, i.e. in our case, a new rule string has been obtained. Another set of rule strings will be generated in this way, some of which will replace previous strings based on fitness selection. If stopping conditions are not met, the conditional probabilities for all nodes in the Bayesian network are updated again using the current set of promising rule strings. Computational results from 52 real data instances demonstrate the success of this approach. It is also suggested that the learning mechanism in the proposed approach might be suitable for other scheduling problems.

An Estimation of Distribution Algorithm for Nurse Scheduling

Schedules can be built in a similar way to a human scheduler by using a set of rules that involve domain knowledge. This paper presents an Estimation of Distribution Algorithm (EDA) for the nurse scheduling problem, which involves choosing a suitable scheduling rule from a set for the assignment of each nurse. Unlike previous work that used Genetic Algorithms (GAs) to implement implicit learning, the learning in the proposed algorithm is explicit, i.e. we identify and mix building blocks directly. The EDA is applied to implement such explicit learning by building a Bayesian network of the joint distribution of solutions. The conditional probability of each variable in the network is computed according to an initial set of promising solutions. Subsequently, each new instance for each variable is generated by using the corresponding conditional probabilities, until all variables have been generated, i.e. in our case, a new rule string has been obtained. Another set of rule strings will be generated in this way, some of which will replace previous strings based on fitness selection. If stopping conditions are not met, the conditional probabilities for all nodes in the Bayesian network are updated again using the current set of promising rule strings. Computational results from 52 real data instances demonstrate the success of this approach. It is also suggested that the learning mechanism in the proposed approach might be suitable for other scheduling problems.

Noise Reduction Technique for a Simulation Optimisation Study

This paper reports on an attempt to apply Genetic Algorithms to the problem of optimising a complex system, through discrete event simulation (Simulation Optimisation), with a view to reducing the noise associated with such a procedure. We are applying this proposed solution approach to our application test bed, a Crossdocking distribution centre, because it provides a good representative of the random and unpredictable behaviour of complex systems i.e. automated machine random failure and the variability of manual order picker skill. It is known that there is noise in the output of discrete event simulation modelling. However, our interest focuses on the effect of noise on the evaluation of the fitness of candidate solutions within the search space, and the development of techniques to handle this noise. The unique quality of our proposed solution approach is we intend to embed a noise reduction technique in our Genetic Algorithm based optimisation procedure, in order for it to be robust enough to handle noise, efficiently estimate suitable fitness function, and produce good quality solutions with minimal computational effort.

'The application of Bayesian Optimization and Classifier Systems in Nurse Scheduling'

Abstract. Two ideas taken from Bayesian optimization and classifier systems are presented for personnel scheduling based on choosing a suitable scheduling rule from a set for each person's assignment. Unlike our previous work of using genetic algorithms whose learning is implicit, the learning in both approaches is explicit, i.e. we are able to identify building blocks directly. To achieve this target, the Bayesian optimization algorithm builds a Bayesian network of the joint probability distribution of the rules used to construct solutions, while the adapted classifier system assigns each rule a strength value that is constantly updated according to its usefulness in the current situation. Computational results from 52 real data instances of nurse scheduling demonstrate the success of both approaches. It is also suggested that the learning mechanism in the proposed approaches might be suitable for other scheduling problems.

'Bayesian Optimisation for Nurse Scheduling'

A Bayesian optimisation algorithm for a nurse scheduling problem is presented, which involves choosing a suitable scheduling rule from a set for each nurse's assignment. When a human scheduler works, he normally builds a schedule systematically following a set of rules. After much practice, the scheduler gradually masters the knowledge of which solution parts go well with others. He can identify good parts and is aware of the solution quality even if the scheduling process is not yet completed, thus having the ability to finish a schedule by using flexible, rather than fixed, rules. In this paper, we design a more human-like scheduling algorithm, by using a Bayesian optimisation algorithm to implement explicit learning from past solutions. A nurse scheduling problem from a UK hospital is used for testing. Unlike our previous work that used Genetic Algorithms to implement implicit learning [1], the learning in the proposed algorithm is explicit, i.e. we identify and mix building blocks directly. The Bayesian optimisation algorithm is applied to implement such explicit learning by building a Bayesian network of the joint distribution of solutions. The conditional probability of each variable in the network is computed according to an initial set of promising solutions. Subsequently, each new instance for each variable is generated by using the corresponding conditional probabilities, until all variables have been generated, i.e. in our case, new rule strings have been obtained. Sets of rule strings are generated in this way, some of which will replace previous strings based on fitness. If stopping conditions are not met, the conditional probabilities for all nodes in the Bayesian network are updated again using the current set of promising rule strings. For clarity, consider the following toy example of scheduling five nurses with two rules (1: random allocation, 2: allocate nurse to low-cost shifts). In the beginning of the search, the probabilities of choosing rule 1 or 2 for each nurse is equal, i.e. 50%. After a few iterations, due to the selection pressure and reinforcement learning, we experience two solution pathways: Because pure low-cost or random allocation produces low quality solutions, either rule 1 is used for the first 2-3 nurses and rule 2 on remainder or vice versa. In essence, Bayesian network learns 'use rule 2 after 2-3x using rule 1' or vice versa. It should be noted that for our and most other scheduling problems, the structure of the network model is known and all variables are fully observed. In this case, the goal of learning is to find the rule values that maximize the likelihood of the training data. Thus, learning can amount to 'counting' in the case of multinomial distributions. For our problem, we use our rules: Random, Cheapest Cost, Best Cover and Balance of Cost and Cover. In more detail, the steps of our Bayesian optimisation algorithm for nurse scheduling are: 1. Set t = 0, and generate an initial population P(0) at random; 2. Use roulette-wheel selection to choose a set of promising rule strings S(t) from P(t); 3. Compute conditional probabilities of each node according to this set of promising solutions; 4. Assign each nurse using roulette-wheel selection based on the rules' conditional probabilities. A set of new rule strings O(t) will be generated in this way; 5. Create a new population P(t+1) by replacing some rule strings from P(t) with O(t), and set t = t+1; 6. If the termination conditions are not met (we use 2000 generations), go to step 2. Computational results from 52 real data instances demonstrate the success of this approach. They also suggest that the learning mechanism in the proposed approach might be suitable for other scheduling problems. Another direction for further research is to see if there is a good constructing sequence for individual data instances, given a fixed nurse scheduling order. If so, the good patterns could be recognized and then extracted as new domain knowledge. Thus, by using this extracted knowledge, we can assign specific rules to the corresponding nurses beforehand, and only schedule the remaining nurses with all available rules, making it possible to reduce the solution space. Acknowledgements The work was funded by the UK Government's major funding agency, Engineering and Physical Sciences Research Council (EPSRC), under grand GR/R92899/01. References [1] Aickelin U, "An Indirect Genetic Algorithm for Set Covering Problems", Journal of the Operational Research Society, 53(10): 1118-1126,